Subtyping Alzheimer's disease and Parkinson's disease using longitudinal electronic health records - PubMed
4 hours ago
- #Disease Subtyping
- #Electronic Health Records
- #Neurodegenerative Diseases
- Study applies transformer-based unsupervised clustering to longitudinal EHR data from over 100,000 patients in UK cohorts to subtype Alzheimer's (AD) and Parkinson's (PD) diseases.
- Identifies five reproducible subtypes for each condition, characterized by distinct comorbidity patterns, symptom trajectories, outcomes, and genetic profiles.
- Highlights a high-mortality AD subtype with motor and cardiovascular features and a genetically susceptible but clinically resilient PD subtype.
- Discovers metabolic-inflammatory and vascular-psychiatric phenotypes shared across AD and PD, suggesting cross-disease mechanisms.
- Provides a scalable framework for early, biologically informed subtyping to support targeted interventions in neurodegenerative diseases.